Is this Continuum under the hood? What are plans for pricing/licensing?

[Lars_E_]

Before I get too far into this...is this Continuum Analytics Python distribution under the hood?  I've been using Continuum 2.4 distribution but I can't get 2.5 distribution to install correctly on Windows 7, there is a menu install error at the end and no human to contact for free distribution.

Also what are licensing/pricing plans?  It's a big jump to go from free distribution to commercial one, so I'd need to know, past Beta, how this thing will be priced and licensed, for development and for end-users of applications.

[Sergey_M_Intel2]

Dear Lars,

No, as of now Intel Distribution for Python does not use Anaconda under the hood. I cannot comment on your bad experience on Windows 7. Let me know if you see it with Intel Python.

Intel Python is available for free. Those who want Premier support for Python may get it thru Intel Parallel Studio license. In other words, if you're Parallel Studio user then you get Premier support for Python at no extra cost.

In particular Intel redistributables within Python go under the same license terms as other Intel libraries.

Hope I answered your questions,

Regards,

Sergey

[Lars_E_]

Hi Sergey,

Can you point me a list of the licensed redistributables such as MKL that are in the bundle and a link to how licensing works for each of those?

You mentioned Continuum in your LinkedIn post so I assumed there was a connection, my mistake.

Also, my understanding is that Python is not thread-safe, so you can't multithread a single Python application into parallel subtasks.  Is that correct?  If so, what are the uses of TBB except to support C++ libraries under the hood that use it, such as MKL?

[Sergey_M_Intel2]

Hi Lars,

EULA is part of Intel Distribution for Python. This is essentially the same license as for other Intel software. Respective section describes conditions under which you can redistribute:

A.    "Redistributables" (if any) are the files listed in the following text files that may be included in the Materials for the applicable Intel Software Development Product: clredist.txt, credist.txt, fredist.txt, redist.txt, and redist-rt.txt.

Since Beta is intended for evaluation purposes we do not allow redistribution in Beta. For final product the redistributables will allow to redistribute Intel MKL, Intel MPI, etc. runtimes under the same conditions as respective standalone products, i.e. conditions won't be more restrictive. Does it make sense?

Regarding thread safety. Python is thread safe; even too thread safe. Python has GIL (Global Interpreter Lock), which serializes essentially every Python operation, disabling scalability. But Python has ThreadPool class, which we recently made TBB-based and some Python packages release GIL. And hence you can execute these modules in parallel threads. For example, NumPy/SciPy release GIL when MKL is called and as a result you get multiple MKL calls running in parallel. But MKL is multi-threaded by itself and this is how you get nested parallelism (which works better with TBB). Does it answer your question?

Regards,

Sergey